Guns in America: National Survey on Private Ownership and Use Of

T O EN F J TM U U.S. Department of Justice R S T A I P C E E D

O J Office of Justice Programs C S F A V M F O I N A C I J S R E BJ G O OJJ DP O F PR National Institute of Justice JUSTICE National Institute of Justice R e s e a r c h i n B r i e f Jeremy Travis, Director May 1997 Issues and Findings Guns in America: National Survey on Discussed in this Brief: Results of a nationally representative tele- Private Ownership and Use of Firearms phone survey (1994) on private ownership and use of firearms by by Philip J. Cook and Jens Ludwig American adults. The survey provides the most complete data available on the private stock of The United States is unique among • Methods of, and reasons for, firearms firearms in the United States. wealthy nations in its vast private inven- acquisition. tory of firearms. The nearly 200 million • Storage and carrying of guns. Key issues: With nearly 200 mil- guns in private hands are used in part for lion guns in private hands, firearms recreation, mostly hunting and target • Defensive use of firearms against crimi- have an important impact on the shooting. But what engenders the most nal attackers. quality of life in America. What is the size and composition of the public controversy over firearms is their Nation's private gun inventory? use against people during either the com- Gun ownership What are the methods of, and rea- mission of or defense against crime. Prevalence. According to conventional sons for, acquiring firearms? How Gun advocates regard firearms as an im- wisdom, about half of American house- are firearms stored? How fre- holds own guns, a belief affirmed by a quently are guns used against portant crime deterrent and source of protection, while control advocates de- long series of national polls dating back criminal attackers? 1 nounce guns for the damage they do in to 1959. Yet data from the 1994 tele- Key findings: The survey data and the hands of criminals. What both groups phone survey (National Survey of Private analysis yielded the following results: can agree on is that widespread owner- Ownership of Firearms—NSPOF) indicate that just 35 percent (plus or minus • In 1994, 44 million Americans ship of firearms has an important impact 1.3 percent) of households own guns. owned 192 million firearms, 65 on the quality of life in America. million of which were handguns. This estimate may be somewhat off the Although there were enough guns To learn more about the role of firearms, mark but not by much. Conventional wis- to have provided every U.S. adult the National Institute of Justice (NIJ) dom appears out of date. with one, only 25 percent of adults sponsored—through a grant to the Police actually owned firearms; 74 per- Foundation—a nationally representative The best available survey series on gun cent of gun owners possessed two telephone survey in 1994 on private own- ownership is the General Social Survey or more. ership and use of firearms by American (GSS), conducted by the National Opin- ion Research Center. Its estimates have • The proportion of American adults (see "Firearms Survey Methodol- ogy"). This Research in Brief reports been lower than some others, in the range households that keep firearms ap- of 40 to 43 percent during the 1990s. In pears to be declining. some of the survey's more important findings, including the following: particular, the GSS estimate for 1994 was • Sixty-eight percent of handgun just 41 percent. Another telephone sur- owners also possessed at least one • Size, composition, and ownership of the vey in 1994 produced a still lower esti- rifle or shotgun. Nation's private gun inventory. mate for gun ownership, 38 percent of households.2 continued… R e s e a r c h i n B r i e f

Issues and Findings Concentration. Despite enough guns in uted across the population, as is evident continued… private hands to provide every adult in from exhibit 2. Most striking is the gen- America with one, only one-quarter of der gap: 42 percent of men but just 9 per- • Gun ownership was highest among middle-aged, college- adults actually own firearms. Those who cent of women owned guns at the time of educated people of rural small- have one gun usually have several: 74 NSPOF. (The gap is even wider when the town America. Whites were percent possessed two or more in 1994. focus is on whether the respondent ever substantially more likely to own owned a gun.) With respect to race, guns than blacks, and blacks more Gun ownership is quite concentrated but whites were substantially more likely to likely than Hispanics. not more so than for other durable goods. own guns than blacks (27 versus 16 per- In marketing circles, the "80/20 rule" cent), and blacks more likely than His- • The most common motivation suggests that the top fifth of all consum- panics (16 versus 11 percent). But for for owning firearms was recre- ers of a product typically account for handguns alone, the ownership rates ation. Forty-six percent possessed a four-fifths of all purchases by value. gun primarily for protection against among blacks and whites were nearly NSPOF data indicate that the top 20 per- crime. equal (13.1 versus 16.5 percent). cent of firearm owners possessed 55 per- • There were 13.7 million firearm cent of privately owned firearms.3 Of gun Gun ownership (and handgun ownership) transactions in 1993–1994, includ- owners in 1994, 10 million individuals was highest among middle-aged,5 college- ing 6.5 million handguns. About owned 105 million guns, while the re- educated people of rural and small-town 60 percent of gun acquisitions in- maining 87 million guns were dispersed America. But one of the best predictors of volved federally licensed dealers. among 34 million other owners. gun ownership was the presence of firearms in the respondent's childhood home. • About 211,000 handguns and Persons owning several guns tended to 382,000 long guns were stolen in People whose parents possessed guns have varied collections, including rifles, noncommercial thefts in 1994. were three times as likely as others to 4 shotguns, and handguns. We find that 68 own one themselves. In fact, 80 percent • Slightly more than half of all pri- percent of handgun owners also owned at of all current gun owners reported that vately owned firearms were stored least one rifle or shotgun, suggesting their parents kept a firearm in the home. unlocked; 16 percent of firearms some experience and interest in the were stored unlocked and loaded. sporting uses of guns. Exhibit 1 provides Motivations. The most common motiva- • In 1994, about 14 million adults additional data on the composition of pri- tion for owning firearms was recreation. (approximately one-third of gun vate gun collections. As shown in exhibit 3, about 35 percent owners) at least once carried a fire- of gun owners (15 million people, 8 per- arm in their vehicles or on their Demographic patterns. In 1994 gun cent of the adult public) hunted in 1994, person for protection. ownership was far from uniformly distrib- and about an equal percentage engaged

• Evidence suggests that this sur- Exhibit 1. Composition of Gun Ownership (1994) a vey and others like it overestimate the frequency with which firearms Own were used by private citizens to de- Handguns Only fend against criminal attack. Own 1 12% 20% Target audience: Criminal justice Own 2+ and public health researchers and 8% practitioners. Legislators and policy- Own Both Own 1 makers at all levels of government. 44% 13%

Own Long Own 2+ Guns Only 23% 36%

a. There were 44 million gun owners in 1994.

2 R e s e a r c h i n B r i e f

Exhibit 2. Gun Ownership Patterns (NSPOF Estimates, 1994)

Total 16 9 25 Male 26 16 42 Sex Female 7 2 9 Percentage Owning Handguns (Whether or 18–24 8816 Not Owning Long Guns As Well) 25–39 Age 16 7 23 xx xx xx 40–64 21 11 32 Percentage 65+ 11 9 20 Owning Long Guns Only Total Percentage White 17 10 27 of Gun Owners Race Black 13 3 16 Hispanic 7411

Less Than High School 10 8 18 Education High School Diploma 15 9 24 More Than High School 19 10 29

Yes 23 13 36 Parents Had Gun in Home? No 7411

0 5 10 15 20 25 30 35 40 45 Percentage of Respondents Owning Guns in sport shooting other than hunting. But most persons do not own guns, and moral," or otherwise objectionable. Given the substantial overlap between the NSPOF included several items to The remaining one-third were at least the two groups, about half (23 million) find out why. In 1994, about two-thirds open to the possibility of obtaining of the Nation's 44 million gun owners of gunless adults were actively opposed firearms and might do so if their finan- participated in a gun sport during to having guns in their homes because cial condition or motivation became 1994. Of those who owned only hand- they viewed guns as dangerous, "im- stronger. For many, the needed moti- guns in 1994, 40 percent used them recreationally, almost entirely for sport Exhibit 3. Recreational Use of Firearms—Percentage of Gun Owners Who shooting other than hunting. Hunt, Do Other Sport Shooting, Do Neither Another reason cited for firearm ownership was self-protection. Overall, 46 percent of gun owners possessed fire- Hunting arms (usually handguns) primarily for Only protection against crime (41 percent 18% for males; 67 percent for females). Al- Neither 48% Both Hunting most three-quarters of those who and Other Sport 17% owned only handguns kept them pri- Shooting marily for self-protection. Of course, some people seek the protection of a 17% gun because they may be dispropor- Sport tionately likely to lead risky lives or Shooting Only 6 (Other Than associate with violent people. Those Hunting) who had been arrested for nontraffic offenses were more likely to own fire- Note: The average number of days hunters said they spent hunting in 1994 was 16.4 days. arms (37 percent compared to 25 per- The average number of days sport shooters said they spent sport shooting in 1994 was cent in the general population). 18.6 days.

3 R e s e a r c h i n B r i e f

Firearms Survey Methodology

T he NIJ-sponsored National Survey Minimums were established for the num- definitions yield figures of 44 and 59 per- of Private Ownership of Firearms (NSPOF) ber of completed interviews with racial cent for the NSPOF. Thus, nonresponse was conducted by Chilton Research Ser- minorities and gun-owning households. bias in our estimates is a real possibility. vices of Drexel Hill, Pennsylvania, during Such households were more likely than Nonetheless, the response rate for this November and December 1994. Data col- others to be included in the final sample. survey is no lower than for other well- lected by the survey were analyzed by the Sampling weights were calculated to ad- executed telephone surveys, and there authors of this Research in Brief. just for this design feature and for other is no reason to believe that this survey used sociodemographic differences between a less representative sample than others.** The telephone survey employed a list- the sample and the U.S. adult population. assisted random-digit-dial sampling Most of the estimates contained in this method, in which every residential tele- Although these adjustments improved Research in Brief rely on the responses of phone number had the same likelihood of the quality of population estimates based those who personally owned firearms. being selected. Each household selected on the NSPOF, some types of estimates The estimates do not rely on the reports in this fashion was scheduled for as many may still be biased. As in every survey, of those who did not personally own a calls as needed (up to a maximum of six) some sample members refused to coop- gun but lived in a gun-owning house- to make contact with the appropriate per- erate and others were never home when hold because our analysis of the NSPOF son and complete the interview. When a the interviewer called. The concern is that data suggests that the survey respond– household was first contacted, the inter- these nonrespondents may tend to differ ents were often unwilling or unable to viewer asked to speak with the adult in from the general population (and the report on guns owned by other adults in the household who had the most recent completed sample) in relevant ways. The the household. For example, we find birthday. Because this method random- scope of that potential problem is usually that in households headed by married izes the selection of respondents from indicated by the response rate. couples, women were much less likely to among the adults living in the household, report a gun in the house (which in most the NSPOF was a probability sample of In the absence of a single accepted defini- cases would belong to their husbands) adults in the United States.* tion of "response rate," two reasonable than were men.

* For details about the GENESYS method employed by Chilton or other survey issues, see Brick, J.M., J. Waksberg, D. Kulp, and A. Starer, "Bias in List- Assisted Telephone Samples," Public Opinion Quarterly, 59:218–235. Also: Waksberg, J., "Sampling Methods for Random Digit Dialing," Journal of the American Statistical Association, 73:40–46, 1978. ** Kleck, G., and M. Gertz, "Armed Resistance to Crime: The Prevalence and Nature of Self-Defense With a Gun," Journal of Criminal Law and Criminol- ogy, 86(1):150–187, Fall 1995. They reported a response rate of 61 percent for their national telephone survey of gun ownership and defensive gun use. In calculating this response rate, they excluded all sample members whom they were unable to contact. By their definition, the NSPOF response rate would be higher than 61 percent.

1994 National Survey of Private Ownership of Firearms (NSPOF) Objectives: Provide national estimates for: • Adult ownership of guns, by gun type. • Sources and motivations for gun acquisition. • Firearm safety and storage. • Defensive use of firearms. • Attitudes toward gun control. Sample: Probability sample of 2,568 noninstitutionalized adults aged 18 and over who are fluent in English or Spanish and live in households with a telephone. Method: Telephone interview with one randomly selected adult from each household. Population estimates: Weighted averages of relevant responses. Standard errors for estimates of population-prevalence rates range up to 1.4 percentage points, somewhat higher for prevalence estimates within subpopulations.

4 R e s e a r c h i n B r i e f vation may have come from an in- Transactions How do people typically acquire fire- creased concern about crime: nearly 5 arms? As shown in exhibit 5, almost all Acquisitions. To date, little informa- percent of respondents reported that guns acquired during 1993 and 1994 tion has been available about gun they planned to obtain a gun for pro- were either purchased by the respondent flows in the United States. The poten- tection against crime within a year. (73 percent) or received as a gift (19 per- tial importance of this information is cent). The remaining 8 percent were ob- its use in evaluating regulation of fire- tained through inheritance, a swap of The stock of guns in private arms commerce. For example, the Gun some kind, or other means. hands Control Act of 1968 restricts interstate The NSPOF-based estimate for the to- shipments to federally licensed firearm dealers (FFLs), who in turn are re- Exhibit 4. Gun Stock Characteristics tal number of privately owned firearms (1994) is 192 million: 65 million handguns, quired to follow laws regulating retail 70 million rifles, 49 million shotguns, transfers. Transactions not involving and 8 million other long guns (exhibit FFLs, known as the "secondary mar- 4A. Estimates of Number of Guns ket," typically do not require 4). Of the handguns, 48 percent were Number in revolvers, 40 percent semiautomatics, recordkeeping and are exempt from the millions Federal requirement (for handguns) of a and 12 percent were reported as "some Handguns other type of handgun" by respondents. waiting period and criminal record Revolvers 31 8 check. Moreover, secondary market Semiautomatics 26 The millions of guns in private hands in- transactions are not subject to regula- Other 8 cluded everything from cheap .22-caliber tory oversight. Thus, knowing the vol- Total 65 "snubbies" to finely made high-powered ume of informal transfers that do or do Rifles rifles worth thousands of dollars. The va- not involve FFLs would be useful. Semiautomatics 28 riety of firearm designs reflects the multi- Other 42 plicity of uses for which they are intended The average firearm in circulation in Total 70 and also influences the weapons' capaci- 1994 was acquired by its present Shotguns 49 ties for harm. Firearm regulations place owner in 1981, with the average hand- Other long guns 8 special restrictions on commerce in short- gun having been acquired in 1983. Total All Guns 192 barreled guns (because they are easily Persons owning handguns in 1994 acquired about 28 percent of them in concealed and disproportionately used in 4B. Magazine Capacity of Handgun 1993–1994, compared with 20 percent crime) and on large-capacity magazines. Stock a of long guns. An estimated 13.7 million From our analysis, we find that the transactions occurred during 1993– Percentage magazine capacity of one-fifth of all 1994, including 6.5 million involving Number of of Handgun handguns was 10 or more rounds (ex- handguns. Sixty percent of long guns Rounds Stock hibit 4B). The barrel of about one in and 68 percent of handguns were new 1–9 rounds 79% six handguns was 3 inches or shorter at the time of acquisition by their 1994 10 or more rounds 21% 7 (exhibit 4C). Comparing handguns ac- owners during the 1993–1994 period. a. The average number of rounds is 8.1. quired in 1993 or 1994 with those acquired prior to 1993 permitted examination of changes in the demand 4C. Length of Barrel for different kinds of handguns over Percentage of Percentage of time. Handguns acquired more re- All Handgun Handguns With cently were more likely to have large Stock Caliber .32 or Under b magazine capacities (37.8 versus 14.1 1–3 inches 17% 37%c c percent held 10 or more rounds) and 4–5 inches 38% 31% c were less likely to be of small caliber, 6 or more inches 45% 38% defined as .32 or under (28.6 versus b. The percentage of all handgun stock having a caliber .32 or under is 34 percent. 38 percent). (See exhibit 4D.) c. These percentages are not of all guns but only of those identified in the middle column.

5 R e s e a r c h i n B r i e f

Exhibit 4. Gun Stock Characteristics (1994) (continued)

4D. Magazine Capacity and Barrel Length by Time of Acquisition Handguns Acquired Handguns Acquired Prior to 1993 in 1993 or 1994 (N=234) (N=91) Percentage of All Percentage of All Magazine Capacity d Handguns Handguns 1–9 rounds 85.9% 62.2% 10 or more rounds 14.1% 37.8%

Percentage of Percentage With Percentage of Percentage With Length of Barrel e Handgun Stock Caliber .32 or Under Handgun Stock Caliber .32 or Under 1–3 inches 17.6% 40.8% 17.4% 33.7% 4–5 inches 35.9% 30.5% 41.6% 31.1% 6 or more inches 46.5% 43.4% 41.0% 22.6% d. The average number of rounds for guns acquired before 1993 is 7.6. For guns acquired in 1993 or 1994 it is 9.5. e. The percentage of all handguns acquired prior to 1993 having a caliber .32 or under is 38 percent. The percentage of all handguns acquired in 1993 or 1994 having a caliber .32 or under is 28.6 percent.

The predominant sources of guns, The average gun obtained in 1993 and were not sure about whether the source unsurprisingly, were stores (60 per- 1994 was worth $392 at the time of was an FFL. Others indicated that the cent). Other important sources in- transfer, with little difference between source was an FFL but then reported cluded family members and handguns and long guns. Fewer than 1 that the transaction was a trade rather acquaintances. The 3 percent of re- in 20 guns acquired during those 2 than a cash sale or that the source was spondents who indicated that they ob- years were valued at less than $100. an acquaintance or family member. If tained guns "through the mail" (which those cases are included, the propor- is illegal for all but FFLs) may have Fifty-seven percent of firearms were tion increases to 64 percent. misremembered or may have referred obtained from stores, pawnshops, or to a mail-order purchase arranged other sources that the respondents were We conclude that approximately 60 through an FFL. certain to have been federally licensed percent of gun acquisitions involved firearm dealers. Some respondents an FFL and hence were subject to

Exhibit 5. Methods and Sources for Gun Acquisition in 1993 and 1994 (NSPOF Estimates) Percentage for Long Guns Percentage for Handguns Percentage for All Guns (N=121) (N=128) (N=251) What Best Describes How You Obtained Your Gun? Bought it 69 77 73 Received it as a gift 22 16 19 Traded something for it 3 2 3 Inherited it 5 4 5 From What Source Did You Obtain This Gun? Gun store 33 55 43 Pawnshop 5 8 6 Other store 18 3 11 Gun show or flea market 4 4 4 Through the mail 3 3 3 Member of the family 22 12 17 Friend or acquaintance 12 13 12 Other 5 3 4

6 R e s e a r c h i n B r i e f

Federal regulations on such matters as younger.11 For each such fatality, there Rifle Association, National Safety out-of-State sales, criminal history are several accidental shootings that Council, or other source. As a group, checks, and recordkeeping. A some- cause serious injury. Guns were also owners who received such training what higher percentage of handgun ac- the means of destruction in 19,590 were no less likely than others to keep quisitions than long gun acquisitions suicides, 210 involving children 14 or guns loaded and unlocked. This sur- involved FFLs. The remaining acquisi- younger. For these reasons, safe han- prising result is consistent with other tions, amounting to about 2 million per dling and storage of firearms have at- recent studies.12 year, were off-the-books transfers in tracted the attention of the public the secondary market. health community. However, a more detailed analysis of NSPOF data that examined the effects Thefts. A major theme highlighted in We found that 20 percent of all gun- of different formal training programs a 1986 survey of incarcerated felons owning households had an unlocked, separately indicated one exception: was that theft was an important means loaded gun in the home at the time of training programs such as those of- of obtaining firearms for those with the survey. This figure was substan- fered by local affiliates of the National criminal intentions: 32 percent of sur- tially higher among handgun-owning Safety Council were associated with a veyed felons had stolen their most re- households than among households significant reduction in the likelihood cently acquired handgun.9 with long guns only—30 percent ver- of keeping a gun unlocked and loaded. sus 7 percent. This result speaks well of that training, Based on the NSPOF, an estimated 0.9 the trainees, or both. percent of all gun-owning households Slightly more than half of firearms of (269,000) experienced the theft of one either type were stored unlocked, but Carrying or more firearms during 1994. About handguns were much more likely to be 211,000 handguns and 382,000 long loaded. Reflecting their predominant Carrying a gun outside the home, es- guns were stolen in noncommercial use in self-defense, handguns were pecially in an urban area, is problem- thefts that year, for a total of 593,000 likely to be stored in bedrooms or ve- atic because the public is at risk if the stolen firearms. Those estimates are hicles of owners or even on their per- carrier is reckless or inclined to vio- subject to considerable sampling error son, while most long guns were kept in lence. For that reason, carrying a fire- but are consistent with earlier esti- gun closets or other out-of-the-way arm in a vehicle or on the person is mates of about half a million guns sto- places (exhibit 6). subject to a variety of State and local len annually.10 regulations. In most States, carrying a Although training programs usually in- concealed gun is prohibited or re- clude suggestions on how to store guns Gun safety stricted to those who have obtained a safely, it does not appear that trainees special license. At the same time, many Gun storage. Of 1,356 accidental are paying attention. More than half States have reacted to public concerns deaths by gunshot in 1994, 185 in- (56 percent) of owners had received about crime by enacting laws under volved children 14 years old and some form of "formal" training from which most citizens can usually obtain a the military, law enforcement, National

Exhibit 6. Storage Method and Location of Firearms (NSPOF Estimates) Percentage for Long Guns Percentage for Handguns Percentage for All Guns (N=437) (N=352) (N=789) Storage Method/Location Gun loaded 11 55 26 Gun loaded and unlocked 7 34 16 Where Gun Kept Bedroom 17 37 24 Gun closet 53 26 44 Other closet 19 11 17 In vehicle or on person 1 16 6 Other 10 8 10

7 R e s e a r c h i n B r i e f concealed-carry permit. Currently, 31 Defensive gun uses home, work, or elsewhere?" Answers States have passed such laws. in the affirmative were followed with NSPOF estimates. Private citizens "How many different times did you use sometimes use their guns to scare off About 14 million adults (approxi- a gun, even if it was not fired, to pro- trespassers and fend off assaults. Such mately one-third of gun owners) car- tect yourself or property in the past 12 defensive gun uses (DGUs) are some- ried firearms for protection at least months?" Negative answers to the first times invoked as a measure of the once during the 12 months preceding DGU question were followed by "Have public benefits of private gun owner- NSPOF. Four million of them indicated you ever used a gun to defend yourself ship. On the basis of National Crime that they carried guns for protection "in or someone else?" (emphasis in origi- Victimization Survey (NCVS) data, one connection with work." Two-thirds who nal). Each respondent who answered would conclude that defensive uses carried guns kept them in their ve- yes to either of these DGU questions are rare indeed, about 108,000 per hicles, while the others sometimes car- was asked a sequence of 30 additional year. But other surveys yield far higher ried them on their person. questions concerning the most recent estimates of the number of DGUs. defensive gun use in which the respond- The occupations of respondents who re- Most notable has been a much publi- ent was involved, including the port carrying guns in connection with cized estimate of 2.5 million DGUs, respondent's actions with the gun, the work are quite diverse. Somewhat sur- based on data from a 1994 telephone location and other circumstances of prisingly, only a quarter of this group survey conducted by Florida State the incident, and the respondent's re- were employed in the protective service University professors Gary Kleck and lationship to the perpetrator. field. The questionnaire does not dis- Mark Gertz.13 The 2.5 million figure tinguish between those who are re- has been picked up by the press and Forty-five respondents reported a de- quired by their employers to carry now appears regularly in newspaper fensive gun use in 1994 against a per- firearms as part of their occupational articles, letters to the editor, editorials, son (exhibit 7). Given the sampling duties and those who do so on their own and even Congressional Research Ser- weights, these respondents constitute initiative. In any event, an estimated 3 vice briefs for public policymakers. 1.6 percent of the sample and repre- million adults who were not in law en- sent 3.1 million adults. Almost half of The NSPOF survey is quite similar to forcement or security carried firearms these respondents reported multiple the Kleck and Gertz instrument and for protection on the job in 1994. DGUs during 1994, which provides provides a basis for replicating their the basis for estimating the 1994 DGU The majority (56 percent) of those who estimate. Each of the respondents in incidence at 23 million. This surpris- carried firearms outside of work did so the NSPOF was asked the question, ing figure is caused in part by a few fewer than 30 days per year, but a sub- "Within the past 12 months, have you respondents reporting large numbers stantial minority (22 percent) rarely yourself used a gun, even if it was not of defensive gun uses during the year; left home without a gun. On any given fired, to protect yourself or someone for example, one woman reported 52! day, 1.1 million people were carrying else, or for the protection of property at guns on their person outside the work- place, while another 2.1 million stored guns in their cars or trucks. Exhibit 7. Defensive Gun Use (DGU) Estimates for 1-Year Recall Period (1994)—Comparison of NSPOF with Kleck and Gertz Estimates Some correlates of gun carrying are NSPOF Estimates Kleck and Gertz worth noting. Males who carried guns 1 All DGUs DGUs Meeting in 1994 were about 2 /2 times as likely Against Kleck and Gertz to have been arrested for a nontraffic Persons Criteria* offense as other men (15 percent ver- 1 Year (N=45) (N=19) (N=66) sus 6 percent). And a disproportionate Estimated number of share of gun carriers resided in the defenders (in millions) 3.1 1.5 2.5 South, where the prevalence of carry- Estimated number of DGUs (In millions) 23.0 4.7 n/a ing guns was almost double that of the rest of the Nation. * In their 1995 DGU study, Kleck and Gertz presented estimates based on only the DGU reports that met certain criteria (see text).

8 R e s e a r c h i n B r i e f

A somewhat more conservative NSPOF sons are conservative, as they assume NSPOF estimates also suggest that estimate is shown in the column of ex- only one defensive gun use per de- 130,000 criminals are wounded or hibit 7 that reflects the application of fender. The results still suggest that killed by civilian gun defenders. That the criteria used by Kleck and Gertz to DGU estimates are far too high. number also appears completely out of identify "genuine" defensive gun uses. line with other, more reliable statistics Respondents were excluded on the ba- For example, in only a small fraction on the number of gunshot cases.14 sis of the most recent DGU description of rape and robbery attempts do vic- for any of the following reasons: the re- tims use guns in self-defense. It does The evidence of bias in the DGU esti- spondent did not see a perpetrator; the not make sense, then, that the NSPOF mates is even stronger when one re- respondent could not state a specific estimate of the number of rapes in calls that the DGU estimates are crime that was involved in the inci- which a woman defended herself with calculated using only the most re- dent; or the respondent did not actu- a gun was more than the total number cently reported DGU incidents of ally display the gun or mention it to of rapes estimated from NCVS (exhibit NSPOF respondents; as noted, about the perpetrator. 8). For other crimes listed in exhibit 8, half of the respondents who reported a the results are almost as absurd: the DGU indicated two or more in the pre- Applying those restrictions leaves 19 NSPOF estimate of DGU robberies is ceding year. Although there are no de- NSPOF respondents (0.8 percent of 36 percent of all NCVS-estimated rob- tails on the circumstances of those the sample), representing 1.5 million beries, while the NSPOF estimate of additional DGUs, presumably they are defensive users. This estimate is di- DGU assaults is 19 percent of all ag- similar to the most recent case and rectly comparable to the well-known gravated assaults. If those percentages provide evidence for additional mil- estimate of Kleck and Gertz, shown in were close to accurate, crime would be lions of violent crimes foiled and per- the last column of exhibit 7. While the a risky business indeed! petrators shot. NSPOF estimate is smaller, it is statis- tically plausible that the difference is due to sampling error. Inclusion of Exhibit 8. Defensive Gun Uses Compared to Total Crime Counts (1994) multiple DGUs reported by half of the 3000 19 NSPOF respondents increases the (In Thousands) estimate to 4.7 million DGUs. 2,478 Some troubling comparisons. If 2500 the DGU numbers are in the right ballpark, millions of attempted assaults, thefts, and break-ins were 2000 foiled by armed citizens during the 12- month period. According to these results, guns are used far more often to 1500 1,299 defend against crime than to perpe- trate crime. (Firearms were used by perpetrators in 1.07 million incidents 1000 of violent crime in 1994, according to NCVS data.) 500 462 466 322 Thus, it is of considerable interest and 316 importance to check the reasonable- ness of the NSPOF estimates before 0 NSPOF NCVS NSPOF NCVS NSPOF NCVS embracing them. Because respondents (DGUs) (Crimes) (DGUs) (Crimes) (DGUs) (Crimes) were asked to describe only their most recent defensive gun use, our compari- Rape and Aggravated AttemptsAssault Robbery

9 R e s e a r c h i n B r i e f

False positives. Regardless of which years but falsely report it as occurring more better? That is doubtful, for two estimates one believes, only a small frac- in the previous year—a phenomenon kinds of reasons: tion of adults have used guns defensively known as "telescoping." in 1994. The only question is whether • First, people who draw their guns to that fraction is 1 in 1,800 (as one would Of course, it is easy to imagine the defend themselves against perceived conclude from the NCVS) or 1 in 100 reasons why that rare respondent who threats are not necessarily innocent (as indicated by the NSPOF estimate actually did use a gun defensively victims; they may have started fights based on Kleck and Gertz's criteria). within the time frame may have de- themselves or they may simply be mis- cided not to report it to the inter- taken about whether the other persons Any estimate of the incidence of a rare viewer. But again, the arithmetic really intended to harm them. Survey event based on screening the general dictates that the false positives will interviewers must take the respondent's population is likely to have a positive likely predominate. word for what happened and why; a bias. The reason can best be explained competent police investigation of the by use of an epidemiological frame- In line with the theory that many DGU same incident would interview all par- work.15 Screening tests are always sub- reports are exaggerated or falsified, we ties before reaching a conclusion. ject to error, whether the "test" is a note that in some of these reports, the medical examination for cancer or an respondents' answers to the followup • Second and more generally, the interview question for DGUs. The er- items are not consistent with respon- number of DGUs tells us little about rors are either "false negatives" or dents' reported DGUs. For example, of the most important effects on crime of "false positives." If the latter tend to the 19 NSPOF respondents meeting widespread gun ownership. When a outnumber the former, the population the more restrictive Kleck and Gertz high percentage of homes, vehicles, prevalence will be exaggerated. DGU criteria (exhibit 7), 6 indicated and even purses contain guns, that that the circumstance of the DGU was presumably has an important effect on The reason this sort of bias can be ex- rape, robbery, or attack—but then re- the behavior of predatory criminals. pected in the case of rare events boils sponded "no" to a subsequent ques- Some may be deterred or diverted to down to a matter of arithmetic. Sup- tion: "Did the perpetrator threaten, other types of crime. Others may pose the true prevalence is 1 in 1,000. attack, or injure you?" change tactics, acquiring a gun them- Then out of every 1,000 respondents, selves or in some other way seeking to only 1 can possibly supply a "false The key explanation for the difference preempt gun use by the intended vic- negative," whereas any of the 999 may between the 108,000 NCVS estimate tim.16 Such consequences presumably provide a "false positive." If even 2 of for the annual number of DGUs and have an important effect on criminal the 999 provide a false positive, the the several million from the surveys victimization rates but are in no way result will be a positive bias—regard- discussed earlier is that NCVS avoids reflected in the DGU count. less of whether the one true positive the false-positive problem by limiting DGU questions to persons who first re- tells the truth. Conclusions ported that they were crime victims. Respondents might falsely provide a Most NCVS respondents never have a The NSPOF provides the most com- positive response to the DGU question chance to answer the DGU question, plete data available on the private for any of a number of reasons: falsely or otherwise. stock of firearms in the United States, including the kinds of guns owned, by • They may want to impress the inter- Unclear benefits and costs from whom they are owned, and for what viewer by their heroism and hence ex- gun uses. Even if one were clever purpose they were acquired. When aggerate a trivial event. enough to design a questionnaire that asked, handgun owners usually gave would weed out error, a problem in in- • They may be genuinely confused due self-protection as their primary motive terpreting the result would remain. for owning guns, while long-gun own- to substance abuse, mental illness, or Should the number of DGUs serve as a simply less-than-accurate memories. ers mentioned hunting or target shoot- measure of the public benefit of pri- ing. Other findings support the • They may actually have used a gun vate gun possession, even in prin- conclusion that handguns are much defensively within the last couple of ciple? When it comes to DGUs, is more likely than long guns to be kept

10 R e s e a r c h i n B r i e f unlocked and ready for use in the on similar surveys. Much debated is home and to be carried in public; they whether the widespread ownership of Quick Access to NIJ are much less likely to be used in firearms deters crime or makes it more Publication News sporting activities. Despite those dif- deadly—or perhaps both—but the ferences, demographic and socioeco- DGU estimates are not informative in F or news about NIJ's most recent nomic patterns of firearm ownership in this regard. publications, including solicitations for 1994 were similar for handguns and grant applications, subscribe to long guns; in fact, most handgun own- For other purposes, the NSPOF is a re- JUSTINFO, the bimonthly newsletter sent to you via e-mail. Here's how: ers also owned one or more long guns. liable reference. Such information is vital to the evaluation of the ongoing • Send an e-mail to [email protected] A fair conclusion is that the more fun- debate over government regulation of damental divide is not between handgun transactions, possession, and use. • Leave the subject line blank. gun and long-gun owners but between • Type subscribe justinfo your name those who own guns and those who do Notes (e.g., subscribe justinfo Jane Doe) in not. Those who like guns, have some the body of the message. 1. For example, the December 1993 experience with them, and have the Or check out the "What's New" sec- Gallup Poll estimated that 49 percent means to obtain them tend to keep sev- tion at the Justice Information Center eral for various purposes. But most of of households possessed a gun. home page: http://www.ncjrs.org. the adult public turns elsewhere for 2. Kleck, G., and M. Gertz, "Armed recreation and protection against crime. Resistance to Crime: The Prevalence linquents, see Sheley, J.F., and J.D. Over time, the relative importance of and Nature of Self-Defense With a Wright, Gun Acquisition and Possession self-protection and sport as motiva- Gun," Journal of Criminal Law and in Selected Juvenile Samples, Research tions for gun acquisition and use has Criminology, 86(1):150–187, Fall 1995. in Brief, Washington, D.C.: U.S. Depart- changed. Perhaps as a result of the in- 3. For a discussion of the "80/20" ment of Justice, National Institute of Jus- creasing urbanization of America, the rule, see Clotfelter, C.T., and P.J. tice, December 1993. For adult felons, overall prevalence of gun ownership Cook, Selling Hope: State Lotteries in see Wright, J.D., and P. Rossi, Armed appears to be declining, as is partici- America, Cambridge, Massachusetts: and Considered Dangerous, Hawthorne, pation in hunting. Proportionately Harvard University Press, 1989. It is New York: Aldine, 1986. fewer households owned firearms in possible that if one could adjust for the 7. Of some interest, given the contro- 1994 than was true in the 1960s and value of guns, the degree of concentra- 1970s, and the younger cohorts are en- versy over the "Saturday Night Spe- tion would be still greater and better cial," is that most short-barreled tering into gun ownership at slower fit the rule. rates than previous ones. When people handguns were not .22 caliber; two- do acquire guns now, the motivation is 4. Fifty-five percent of all individuals thirds were in excess of .32 caliber, the more likely self-defense than in the who owned four or more guns had at least same as for longer barreled handguns. past: The mix of new firearms sold in one in each of these three categories. 8. Cook, P.J., S. Molliconi, and T.B. 1994 was equally divided between Cole, "Regulating Gun Markets," handguns and long guns, whereas 25 5. The NSPOF offers evidence that gun ownership is declining. Not only Journal of Criminal Law and Crimi- years earlier twice as many long guns nology, 86(1):59–92, Fall 1995. were sold.17 were middle-aged people more likely to own a gun in 1994 than those under 9. Wright and Rossi, 1986, 183. The NSPOF does not provide much age 40, but they were also more likely evidence on whether consumers who to have acquired a gun by age 21. 10. The standard error for this point buy guns for protection against crime estimate is about four-tenths of a per- get their money's worth. The NSPOF- 6. Surveys of juvenile delinquents and centage point. Thus, the 95-percent based estimate of millions of DGUs adult felons confirm the importance of confidence interval ranges from each year greatly exaggerates the true self-defense as a motive for gun posses- around 0.1 percent to 1.7 percent of number, as do other estimates based sion by active criminals. For juvenile de- gun-owning households. Cook,

11 Molliconi, and Cole, 1995, use data from the National Crime Victimization Philip J. Cook, Ph.D., is ITT/ length report, as submitted to NIJ Survey for the period 1987–1992 to Sanford professor of public policy by the authors and on which this estimate 511,000 stolen guns per studies, Duke University, and Jens Research in Brief is based, may be year. Ludwig, Ph.D., is assistant profes- obtained through interlibrary loan or, for a fee, as a photocopy from 11. Monthly Vital Statistics Report, sor of public policy, Georgetown the National Criminal Justice Ref- Department of Health and Human University. erence Service (call 800–851–3420 Services, 45(3S), September 30, 1996. The national survey was supported for cost and other information). Table 16. under grant number 93–IJ–CX– The Police Foundation has pub- 12. Hemenway, D., S.J. Solnick, and 0017, awarded to the Police Foun- lished Guns in America: Results of D. Azrael, "Firearm Training and Stor- dation by the National Institute of a Comprehensive National Survey age," Journal of the American Medical Justice, Office of Justice Programs, on Firearms Ownership and Use, Association, 273(1):46–50, 1995. U.S. Department of Justice. which is based on the full-length 13. Kleck and Gertz, 1995. This Research in Brief is available report submitted to NIJ by the online at the Justice Information authors. For ordering information, 14. In 1994 about 17,000 people were Center home page (http:// call 202–833–1460 or write to the shot dead in criminal assaults and www.ncjrs.org) and available Police Foundation, Attn: Publications, justifiable homicides. Given what we through fax-on-demand (800–851– 1001 22nd Street N.W., Washing- know about the case fatality rate, 3420 or 301–251–5518; ask for ton, D.C. 20037. fewer than 100,000 nonfatal gunshot document number 1026). The full- woundings were known to the police. (See Cook, P.J., "The Case of the Missing Victims," Journal of Quantita- 16. Cook, P.J., "The Technology of Findings and conclusions of the research re- tive Criminology, 1985). Presumably, Personal Violence." In M. Tonry (ed.), ported here are those of the authors and do not necessarily reflect the official position or poli- the true number of justifiable shootings Crime and Justice: A Review of Re- cies of the U.S. Department of Justice. was just a fraction of this total. search (Vol. 14), Chicago: University of Chicago Press, 1–71, 1991. 15. Hemenway, D., "Survey Research The National Institute of Justice is a com- and Self-defense Gun Use: An Expla- 17. Cook, P.J., "Notes on the Avail- ponent of the Office of Justice Programs, which also includes the Bureau of Justice nation of Extreme Overestimates," ability and Prevalence of Firearms," Assistance, Bureau of Justice Statistics, Harvard Injury Control Center Dis- American Journal of Preventive Medi- Office of Juvenile Justice Delinquency cussion Paper, 1996. cine, 9 Supp:33–38, May/June 1993. Prevention, and the Office for Victims of Crime.

NCJ 165476

U.S. Department of Justice BULK RATE Office of Justice Programs POSTAGE & FEES PAID National Institute of Justice DOJ/NIJ Permit No. G–91 Washington, DC 20531

Official Business Penalty for Private Use $300